PRaT: "Pace Rhythm and Timing" or "PRetentious audiophile Trash"?

[jkeny]

I started a thread here a while ago about Analogue playback wander & digital jitter
http://www.whatsbestforum.com/showthread.php?23126-Analogue-playback-Wander&highlight=jitter

 

[Atmasphere]

This makes sense to me. Do you believe the negative effects of aliasing are improved by better clock rate? As said, I feel clock progression and PRaT improvement have closely correlated over the years.

If by better clock rate, you mean 'higher' than, yes, its easier to reduce aliasing if the clock speed is high enough so that the input filter is effective at preventing HF harmonics from causing aliasing.

One thing to note, digital audio recording do not seem to suffer from issues that we are discussing - the problems seem to be in the playback stage.
We can playback old digital recordings on a good digital audio system & experience the full PRAT & liveness of the recording

Sorry- This statement is false. Older digital recording equipment had a terrible time with aliasing problems. About 30 years ago I made a recording of a 20-20KHz sweep tone, rising at a fairly slow rate. The accompanying 'birdies' cased by aliasing were easy to hear. The newer digital recording gear does a much better job. If you can record a sweep tone like that (from an analog generator) without birdies then we're making some progress.

 

[jkeny]

If by better clock rate, you mean 'higher' than, yes, its easier to reduce aliasing if the clock speed is high enough so that the input filter is effective at preventing HF harmonics from causing aliasing.

I don't think he means a higher speed clock - I understand it to mean lower jitter clock (usually considered better & more expensive)

If I can try to answer his question? In one word - no! aliasing has nothing to do with clock quality

Sorry- This statement is false. Older digital recording equipment had a terrible time with aliasing problems. About 30 years ago I made a recording of a 20-20KHz sweep tone, rising at a fairly slow rate. The accompanying 'birdies' cased by aliasing were easy to hear. The newer digital recording gear does a much better job. If you can record a sweep tone like that (from an analog generator) without birdies then we're making some progress.

Ok, I'm not talking about recordings which have aliasing issues - they are obviously broken, by definition & break the Shannon-Whittaker basic premise of a band-limited signal. if this is not achieved then all bets are off in digital audio.
I'm not even sure what you are talking about - you did recordings with recording equipment which had aliasing problems? Can you name these devices? I would be surprised if any digital recording was released with aliasing problems as it is such a simple thing to avoid?? Are you saying you found such recordings - please provide details?

 

[RogerD]

One thing to note, digital audio recording do not seem to suffer from issues that we are discussing - the problems seem to be in the playback stage.
We can playback old digital recordings on a good digital audio system & experience the full PRAT & liveness of the recording

Yes, I'm saying that our tests are not revealing what our ears are telling us & a lot of it, IMO, has to do with how digital audio is handling dynamic signals, whether the reference voltage fluctuates i.e does the ground noise change with signal processing

You can see on many. many DAc datasheets that the noise floor changes with amplitude. This is ignored because it's considered that the noise floor is low enough to be inaudible. But this assumption leaves out some aspects which may be important - when handling dynamic, overlaid frequencies at different amplitudes what is the noise floor behaviour in this scenario - is it additive - ie.e the same as testing with (30 - 40) multitone signals shows an additive change in the noise floor much higher than the IMD test using 19KHz + 20KHz tones.

The second aspect to consider is just how sensitive are we to a fluctuating noise floor? We know that our auditory processing can actually ignore a fairly high level noise floor once it is fixed in amplitude - hence vinyl & analogue playback has been so successful but what about fluctuating noise floors? We need to delve into auditory processing for some understanding of this. Our auditory processing systems are continually analysing the signals being delivered on the auditory nerve. At a high level we form foreground & background auditory planes similar to the way we do the same in visual perception. So the room noise/reverberations is the background container in which the music is being played. Our auditory models of the real acoustic world don't expect this auditory background plane to change so noise fluctuations subconsciously warn us that something about the room has changed. How this effects us at various levels can be discussed

What happens when you remove the current noise from the digital front end? A increase in PRAT...clarity,dynamics,and overall musicality. John..you are right redbook is better when this is done. Bits are bits.

 

[Sencha]

I love PRAT also.

Do Lynn and Naim bring the PRAT out from the music or do they "engineer" it?

Naim were at least (don't know about now), always keen to state the transient response time of their amplifiers, something that indicated to me that they thought this was an important part of amplifier design, at least to get the sound they were looking for. They also always (back in the day) only ever had single output transistors which they claimed improved the timing of an amplifier albeit reducing its potential power output. Another biggie for them was for owners to partner up a Naim power amp with a Naim pre-amp. Other than helping them to sell more pre-amps it also set the frequency band in which the power amplifier operated in. Thus the power amp would not be asked to reproduce signals it was not comfortable doing. This all helped to keep the amplifier in its sweet spot and enabled it to be able to stop and start amplifying signal as requested.

The Linn LP12 has gone through many incarnations now but still managed to retain its relationship to PRaT. Obviously when designing products the design engineers will try several variants of any component until he or she is satisfied the results are what they want to achieve. Whether they knew how the turntable was going to sound like when they first produced it (Linn were using big American amps at the time but I'm unsure of the make) but once partnered with a Naim amp (Nap 160 at the time I believe) a marriage was made and a whole new doctrine/religion on PRaT was born.

It has obviously been at least a semi important aspect of music reproduction as here we are 45 years later and still discussing its merits in replay. To hang my colours to the mast I have to tell you I like it. I ran a Linn/Naim system for many years and still love that sound, however there are plenty who don't like it at all and also equipment manufacturers who manage to impart the feeling of PRaT in the music and give a whole lot more as well. Once you have been used to a certain thing though it can be hard to move away given that there will be a lot you love about that certain type of musical reproduction. If you have never heard a good, and that is the caveat......good, Linn/Naim system then it is well worth a listen especially the older olive or chrome bumper series of Naim amps. I hope many of you will get the chance to hear they good aspects of what they offer the art of musical reproduction even if that ultimately is not want moves or excites you. In many other aspects of replay there are far better systems available in the modern era.

 

[jkeny]

What happens when you remove the current noise from the digital front end? A increase in PRAT...clarity,dynamics,and overall musicality. John..you are right redbook is better when this is done. Bits are bits.

Yes, Roger in systems which interconnect a number of different components (& most all of our systems do) the ground has to be carefully considered (as it does within audio devices).
With the introduction of the computer into this chain, ground issues & effects became very much magnified
A lot of what we are involved in is ameliorating the ground effects that often result from interconnecting these devices - so your strapping together of the chassis of devices in your system is one way to provide alternate paths for stray currents (usually current leakage issues from SMPS) which would normally infect our signal ground & result in meh sound with depressed clarity, dynamics & PRAT. As you say RBCD sounds wonderfully dynamic & high-res is not something that is needed

I address it by isolating the computer from it's downstream devices which also addresses this but both approaches can be additive

 

[jkeny]

Naim were at least (don't know about now), always keen to state the transient response time of their amplifiers, something that indicated to me that they thought this was an important part of amplifier design, at least to get the sound they were looking for. They also always (back in the day) only ever had single output transistors which they claimed improved the timing of an amplifier albeit reducing its potential power output. Another biggie for them was for owners to partner up a Naim power amp with a Naim pre-amp. Other than helping them to sell more pre-amps it also set the frequency band in which the power amplifier operated in. Thus the power amp would not be asked to reproduce signals it was not comfortable doing. This all helped to keep the amplifier in its sweet spot and enabled it to be able to stop and start amplifying signal as requested.

The Linn LP12 has gone through many incarnations now but still managed to retain its relationship to PRaT. Obviously when designing products the design engineers will try several variants of any component until he or she is satisfied the results are what they want to achieve. Whether they knew how the turntable was going to sound like when they first produced it (Linn were using big American amps at the time but I'm unsure of the make) but once partnered with a Naim amp (Nap 160 at the time I believe) a marriage was made and a whole new doctrine/religion on PRaT was born.

It has obviously been at least a semi important aspect of music reproduction as here we are 45 years later and still discussing its merits in replay. To hang my colours to the mast I have to tell you I like it. I ran a Linn/Naim system for many years and still love that sound, however there are plenty who don't like it at all and also equipment manufacturers who manage to impart the feeling of PRaT in the music and give a whole lot more as well. Once you have been used to a certain thing though it can be hard to move away given that there will be a lot you love about that certain type of musical reproduction. If you have never heard a good, and that is the caveat......good, Linn/Naim system then it is well worth a listen especially the older olive or chrome bumper series of Naim amps. I hope many of you will get the chance to hear they good aspects of what they offer the art of musical reproduction even if that ultimately is not want moves or excites you. In many other aspects of replay there are far better systems available in the modern era.

I'm not intimate with the Naim sound but I believe this is where PRAT originated. But from what I hear they sacrificed other aspects like soundstage & imaging to focus on PRAT.

To me this seems mistaken as when a system begins to achieve a certain level of performance, PRAT will automatically emerge as will soundstage & imaging

 

[RogerD]

One thing to note, digital audio recording do not seem to suffer from issues that we are discussing - the problems seem to be in the playback stage.
We can playback old digital recordings on a good digital audio system & experience the full PRAT & liveness of the recording

Yes, I'm saying that our tests are not revealing what our ears are telling us & a lot of it, IMO, has to do with how digital audio is handling dynamic signals, whether the reference voltage fluctuates i.e does the ground noise change with signal processing

You can see on many. many DAc datasheets that the noise floor changes with amplitude. This is ignored because it's considered that the noise floor is low enough to be inaudible. But this assumption leaves out some aspects which may be important - when handling dynamic, overlaid frequencies at different amplitudes what is the noise floor behaviour in this scenario - is it additive - ie.e the same as testing with (30 - 40) multitone signals shows an additive change in the noise floor much higher than the IMD test using 19KHz + 20KHz tones.

The second aspect to consider is just how sensitive are we to a fluctuating noise floor? We know that our auditory processing can actually ignore a fairly high level noise floor once it is fixed in amplitude - hence vinyl & analogue playback has been so successful but what about fluctuating noise floors? We need to delve into auditory processing for some understanding of this. Our auditory processing systems are continually analysing the signals being delivered on the auditory nerve. At a high level we form foreground & background auditory planes similar to the way we do the same in visual perception. So the room noise/reverberations is the background container in which the music is being played. Our auditory models of the real acoustic world don't expect this auditory background plane to change so noise fluctuations subconsciously warn us that something about the room has changed. How this effects us at various levels can be discussed

Yes, Roger in systems which interconnect a number of different components (& most all of our systems do) the ground has to be carefully considered (as it does within audio devices).
With the introduction of the computer into this chain, ground issues & effects became very much magnified
A lot of what we are involved in is ameliorating the ground effects that often result from interconnecting these devices - so your strapping together of the chassis of devices in your system is one way to provide alternate paths from stray currents (usually current leakage issues from SMPS) which would normally infect our signal ground & result in meh sound with depressed clarity, dynamics & PRAT. As you say RBCD sounds wonderfully dynamic & high-res is not something that is needed

I address it by isolating the computer from it's downstream devices which brings the same results

Yes, by removing 99 pct of the noise, it is very possible to achieve excellent PRAT and even a multi dimensional soundstage.

 

[microstrip]

If by better clock rate, you mean 'higher' than, yes, its easier to reduce aliasing if the clock speed is high enough so that the input filter is effective at preventing HF harmonics from causing aliasing.



Sorry- This statement is false. Older digital recording equipment had a terrible time with aliasing problems. About 30 years ago I made a recording of a 20-20KHz sweep tone, rising at a fairly slow rate. The accompanying 'birdies' cased by aliasing were easy to hear. The newer digital recording gear does a much better job. If you can record a sweep tone like that (from an analog generator) without birdies then we're making some progress.

I think we are now mixing two very different thinks - sampling rate and clock phase noise. Which are you addressing?

 

[morricab]

I am trying to understand the term PRaT so that police wouldn’t arrest me when I misuse it.

So PRaT can only be used with tt? Can it be used with other equipments?

Kind regards,
Tang

No, it was an example only. However, it also shows that a company was aware that people like a pacey sound and deliberately ran the TT a couple % too fast.

I heard a particular lack of PRaT with late 90s single bit CD players and Dacs. This was possibly tied to jitter and the shaping algorithms that made 1 bit possible.

I have heard it also with some amps although the reasons there are not so clear but phase shifts might be a culprit.

As to Lynn and Naim, well their Class B, high negative feedback, amps are real effective at capturing the basic rhythm and beat of the music because they have stripped away all subtlety and low level information so the first order sounds become quite prominent at the expense of the nuance of music. The result is pretty dry sounding but arguably with PRaT. The real trick is to get PRaT without sacrificing nuance.

 

[morricab]

That is my experience as well, see earlier in the discussion.



The interesting thing is, why is analog naturally so good at this, while absolute speed variations in the medium are much greater than on digital with its picosecond jitter? I assume this must have to do something with the lockstep of signals within the entire frequency range relative to one another, even if in macro-terms the speed of the turntable (or analog tape) fluctuates. Or perhaps the problems in digital also have to do with, as John suggested in post #10, noise-shaping that fluctuates in correlation with the signal.

Analog is not immune. A lot of belt drives suffer to some degree. DDs tend to have better PRaT but often dry out the sound (micro speed variations?). A good DD has PRaT and nuance...so does a good belter.

Modern digital suffers still as well, particularly delta/sigma based DACs.

 

[morricab]

This is indeed the great mystery. To me, I often wonder if it's not about the actual speed of the music replication (i.e., the speed of the TT or the Digital playback) but the way the waves are stitched together. Even if a vinyl record was printed from a digital file it can have more PRaT to me than the digital playback of the same file. I have often wondered if its because the TT system itself is a mechanical device (just like a like a musical instrument) that generates a continuous (analog) signal from a continuous mechanical movement as opposed to a digital file of 1"s and 0's stitched together by a clock and processor. No matter how good clock and processor are, you still have femtosecond disturbances the brain can detect as the gatekeeper of the limbic system that in turn causes the brain to send the activity to the cortex.

Every sound in nature (that resulted in the human brain's evolution over the last 2 million plus years) was generated from a mechanical action in the physical world which vibrates air waves. Even though in digital playback the air is vibrated by a mechanical action (the speaker cone), this action comes from reconstruction of the original natural mechanical action using a clock and processor that by definition do not create a perfectly continuous wave (no matter how good the clock and processor). Does the brain somehow have the ability to discern this and send the signals that it can't figure out to the cortex where it tries to think about them logically as opposed to sending the signals to the limbic system where their processing would generate emotional pleasure?

I am of course talking out my bum here, but I have never seen a compelling explanation for these differences.

IMO it can tell. There is a reason that even picoseconds can be detected. Evolution has fine tuned the system over a very long time where digital was nowhere to be found. Bits at the wrong time create small bits of cognitive dissonance that gives away the game that it's fake. Patterns matter more than individual bits and when the pattern doesn't quite fit then it is perceived as not really based on the imprint of real sounds you have heard since birth.

Now I think if you did a cruel experiment where you exposed a baby to only digitally synthesized sounds then perhaps that child would not be able to tell the difference because the digital pattern is being set as normal.

 

[morricab]

Yes, I'm saying that our tests are not revealing what our ears are telling us & a lot of it, IMO, has to do with how digital audio is handling dynamic signals, whether the reference voltage fluctuates i.e does the ground noise change with signal processing

You can see on many. many DAc datasheets that the noise floor changes with amplitude. This is ignored because it's considered that the noise floor is low enough to be inaudible. But this assumption leaves out some aspects which may be important - when handling dynamic, overlaid frequencies at different amplitudes what is the noise floor behaviour in this scenario - is it additive - ie.e the same as testing with (30 - 40) multitone signals shows an additive change in the noise floor much higher than the IMD test using 19KHz + 20KHz tones.

The second aspect to consider is just how sensitive are we to a fluctuating noise floor? We know that our auditory processing can actually ignore a fairly high level noise floor once it is fixed in amplitude - hence vinyl & analogue playback has been so successful but what about fluctuating noise floors? We need to delve into auditory processing for some understanding of this. Our auditory processing systems are continually analysing the signals being delivered on the auditory nerve. At a high level we form foreground & background auditory planes similar to the way we do the same in visual perception. So the room noise/reverberations is the background container in which the music is being played. Our auditory models of the real acoustic world don't expect this auditory background plane to change so noise fluctuations subconsciously warn us that something about the room has changed. How this effects us at various levels can be discussed

A fluctuating noise floor is indeed a problem, particularly when it is a signal correlated noise floor. This will mask low level information and affect the perception of a sound as real. This is also an issue with amps using negative feedback.

 

[morricab]

I have tried to introduce this subject to debate more than once, but was quickly forgotten. All digital data has intrinsic jitter, most of the time of much higher magnitude than our current DACs. However no one speaks about it.

Digital data itself does not have jitter. Jitter is a timing error in the reading of a particular bit. it is a dynamic error in the playback system.

 

[Legolas]

That is my experience as well, see earlier in the discussion.



The interesting thing is, why is analog naturally so good at this, while absolute speed variations in the medium are much greater than on digital with its picosecond jitter? I assume this must have to do something with the lockstep of signals within the entire frequency range relative to one another, even if in macro-terms the speed of the turntable (or analog tape) fluctuates. Or perhaps the problems in digital also have to do with, as John suggested in post #10, noise-shaping that fluctuates in correlation with the signal.

Agree there. Noise shaping in many modern DACs to get the perceived perfect lab report, that has killed music for years IMO. They seem to have more and more complex filtering and circuits designs but 25 years on many DACs just don't do it for me, no PRaT, in fact pretty lifeless. But hey, we have a nice lab report.

In digital, half of the enjoyment is the timing and music pace / sway / foot tapping. And to enjoy that I think we have to have zero noise and digital hash in the sound, along with dynamics and drive of course.

Why do we 'accept' tape hiss and still enjoy the music? My many early years on cassette tapes attest to that. Why do we accept pops and clicks in a TT replay system? It seems those things in our brains we have no problem with, put it to the back and not analyse it. Maybe like wind noise or a distant bird or something.

But digital noise or hash, especially inconsistent noise, speaking for myself, I find it gets my attention (unwanted attention). It seems like a piece of plastic in a field, stands out right away, ruins the view (the sound). Bit like a mobile phone going off at the other end of a concert hall, it bugs me and ruins the illusion, reminds me I am listing to a 'digital' device.

It may be the DACs at fault here, but some of the blame is also IMO squarely at the original studio where it was recorded / mastered. Adele for example, terribly recorded masters and over the top processing to get her voice to sound more pitchy. Why did they do that? Totally unnecessary, her voice sounds amazing without it. Interestingly there is a big forum where studio engineers post a lot, and Adel is on the 'bad' list, along with Metallica Death Magnetic, many other modern studio disastrous recordings.

So I think there is noise and noise. Some noise we perceive as a 'natural' element we are ok with, no problem, it isn't bothering me. Other noise we perceive as unnatural, we have an issue with.... and right there we have even coined a phrase for it - 'digital sound'. Oh my.....

My current DAC is the best and most natural sound I have heard in digital to date.

Thoughts?

 

[Al M.]

Analog is not immune. A lot of belt drives suffer to some degree. DDs tend to have better PRaT but often dry out the sound (micro speed variations?). A good DD has PRaT and nuance...so does a good belter.

Ok, perhaps I am spoiled by having been only exposed to good analog lately.

Modern digital suffers still as well, particularly delta/sigma based DACs.

That depends on the implementation. The Berkeley Alpha DAC 2, which is delta/sigma, is excellent in PRaT. It can really rock and swing. It only shows some weaknesses in direct comparison with my current DAC, the Schiit Yggdrasil (ladder multibit), which is rhythmically a stunner.

 

[Al M.]

I heard a particular lack of PRaT with late 90s single bit CD players and Dacs. This was possibly tied to jitter and the shaping algorithms that made 1 bit possible.

I heard that in the early 90s with those single bit CD players and Dacs, but basically we agree, those single bit ('bitstream') players were the worst offenders, especially the cheap ones (my Meridian bitstream players, the 208 and the 602/606, were decent in PRaT).

As to Lynn and Naim, well their Class B, high negative feedback, amps are real effective at capturing the basic rhythm and beat of the music because they have stripped away all subtlety and low level information so the first order sounds become quite prominent at the expense of the nuance of music.

Didn't know that about Linn amps. Interestingly, in digital they went the other way. The Linn Numerik A/D converter was very good at low-level signals, and I have some really great Mozart string quartet recordings with it.

 

[jkeny]

IMO it can tell. There is a reason that even picoseconds can be detected. Evolution has fine tuned the system over a very long time where digital was nowhere to be found. Bits at the wrong time create small bits of cognitive dissonance that gives away the game that it's fake. Patterns matter more than individual bits and when the pattern doesn't quite fit then it is perceived as not really based on the imprint of real sounds you have heard since birth.

Now I think if you did a cruel experiment where you exposed a baby to only digitally synthesized sounds then perhaps that child would not be able to tell the difference because the digital pattern is being set as normal.

Yes, consider auditory perception as both a learning machine & analysis machine - it learns from what we hear everyday & analyses & extracts elements into auditory behaviour models to which we compare & analyse what we hear.

A recent article in NYT https://www.nytimes.com/2018/03/06/...on=inside-nyt-region&WT.nav=inside-nyt-region should bring clarity for some that auditory & visual perception use a model-based approach. In this experiment the pinna was changed in shape which affects our ability to spatially locate the source of sounds. Over about a week the wearer learned to interpret the new pattern of signals & the source locating ability returned. After this week the silicon inserts were removed & the auditory locating abilities quickly returned. So the 'old' internal model which conferred source location abilities became almost instantly accessible. The same experiments were done years ago with using prism eyeglasses which inverted the image falling on the Fovea (the normal image on the Fovea is actually upside down). The wearers took about a week of adjustment before the saw the world upright. The fact that we adapt to changes in the eyes or ears which then give us the same result as our previous perception gave, means that we learn/lay down new aspects/models by correlation between our senses & our understanding of how the world works within that perceptual framework. So prism glasses that flip the image falling on the Fovea into the correct, upright orientation is disorienting at first but after about a week the visual world is no longer perceived as flipped but is correctly orientated.

And here's the interesting bit - some weeks or so later if we put back the silicon inserts in the ear we almost immediately adjust & our auditory source location abilities are the same as after the week's training. The same applies to refitting the prism glasses - we quickly readjust & again perceive the image in the correct orientation.

So we learn the perceptual patterns that represent the world as we understand it.

Understanding this about how our perceptions work, why do some people dismiss the idea of long term listening being useless? Does/could auditory perception not also build a similar (probably weaker) model of the sound signature of our system over time & introducing a change to this sound be perceivable?

 

[Al M.]

Yes, consider auditory perception as both a learning machine & analysis machine - it learns from what we hear everyday & analyses & extracts elements into auditory behaviour models to which we compare & analyse what we hear.

A recent article in NYT https://www.nytimes.com/2018/03/06/...on=inside-nyt-region&WT.nav=inside-nyt-region should bring clarity for some that auditory & visual perception use a model-based approach. In this experiment the pinna was changed in shape which affects our ability to spatially locate the source of sounds. Over about a week the wearer learned to interpret the new pattern of signals & the source locating ability returned. After this week the silicon inserts were removed & the auditory locating abilities quickly returned. So the 'old' internal model which conferred source location abilities became almost instantly accessible. The same experiments were done years ago with using prism eyeglasses which inverted the image falling on the Fovea (the normal image on the Fovea is actually upside down). The wearers took about a week of adjustment before the saw the world upright. The fact that we adapt to changes in the eyes or ears which then give us the same result as our previous perception gave, means that we learn/lay down new aspects/models by correlation between our senses & our understanding of how the world works within that perceptual framework. So prism glasses that flip the image falling on the Fovea into the correct, upright orientation is disorienting at first but after about a week the visual world is no longer perceived as flipped but is correctly orientated.

And here's the interesting bit - some weeks or so later if we put back the silicon inserts in the ear we almost immediately adjust & our auditory source location abilities are the same as after the week's training. The same applies to refitting the prism glasses - we quickly readjust & again perceive the image in the correct orientation.

So we learn the perceptual patterns that represent the world as we understand it.

Very interesting.

Understanding this about how our perceptions work, why do some people dismiss the idea of long term listening being useless? Does/could auditory perception not also build a similar (probably weaker) model of the sound signature of our system over time & introducing a change to this sound be perceivable?

I agree. Long term listening is also uniquely apt, unlike short term listening, to teach us what to listen for. Sometimes certain differences between components are immediately obvious, while other, perhaps just as important ones, only are discovered upon prolonged exposure.

In that context it is interesting how the initial reaction of seasoned reviewers of Stereophile (even Dick Olsher!) was all rave about the first CD players in 1983/84. Only later on they became highly critical of the medium, after learning to listen to the weaknesses of digital once the novelty, and the attention to some strengths of CD, had worn off.

 

[jkeny]

IMO it can tell. There is a reason that even picoseconds can be detected. Evolution has fine tuned the system over a very long time where digital was nowhere to be found. Bits at the wrong time create small bits of cognitive dissonance that gives away the game that it's fake. Patterns matter more than individual bits and when the pattern doesn't quite fit then it is perceived as not really based on the imprint of real sounds you have heard since birth.

Now I think if you did a cruel experiment where you exposed a baby to only digitally synthesized sounds then perhaps that child would not be able to tell the difference because the digital pattern is being set as normal.

I believe we are not looking at the correct jitter figure - the picoseconds/femtoseconds is actually a measure of a range of frequencies which avoids close-in jitter measurement or close-in phase noise as it is more commonly referred to. So the phase noise measured close into the fundamental is actually far higher than the >-150dB measured at frequencies > 1KHz (you will find this is the lower freq range used in pico/femto second jitter figures).

Here's a plot of the phase noise of a good clock down to 0.1Hz (this is not often given for clocks as it is difficult to measure & not very flattering for most clocks)
As can be seen in this good clock plot the phase noise is rising from about -145dB @ 1KHz to about -25dB @ 0.01Hz - were not talking about anything like pico or femto second jitter in this < 10Hz frequency range



I have tested these low phase noise clocks & they bring an additional clarity & solidity to the reproduced music
I believe this is where the real improvement is to be found & 'better' clocks should have lower close-in phase noise - forget about pico/femto

I'm not the only one thinking along these lines - see here http://www.customanalogue.com/terra_firma.htm
And these FFTs are illustrative

The second FFT showing the power in frequencies closer to the fundamental.As stated on that website:

the type of jitter discussed below has been measured/confirmed by Paul Miller in the UK, with access to equipment with 100 times the normal sensitivity and this magnification aimed at knowing where to look. Normal jitter does not cause digital sound (more like coluration we hear and adjust to in speakers). But this is a specific ultra-low level type of jitter not yet fully recognised for what it is and what it does.

In the DUAL graph in the first graph, we can see that there was very little difference between Red and Black, but it was also stated that Black was observed to sound quite a bit better than Red.

Now what was contained in the Blue box has been magnified something like 100x - and NOW we can see a significant different. UNL Jitter is very obvious. This was a measurement made by some extreme piece of measurement worth hundreds of thousand dollars (by Paul Miller).

What does close-in phase noise mean? It means that if reproducing a 1KHz tone there are frequency fluctuations in this 1Khz tone - so at 1Hz away, these errors will result in a 1001Hz tone at about -85dB down; at .1Hz away from the tone these errors are only -55dB down & closer in -20dB down. So the tone is fluctuating about it's pure signal. You could see this on FFTs as a broadening of the base of the tone if the frequency X-axis was fine enough so that 1Hz away from the fundamental was visible. But there are many reasons why we are not seeing this - people don't subscribe to the above logic - poo pooing that < 1Hz jitter can be perceptible; it is difficult to do these close-in measurements & takes time; FFTs can broaden the base anyway depending on what windowing method is employed.

So what is the perceptual effect of slightly fluctuating frequencies? Perhaps it's a loss of clarity, soundstage definition & layering? One of the interesting things you see reported by many people is that the first thing they notice with low jitter is the sound quality of the bass which surprises them, expecting HF to be most affected. It could be that a small error in the frequency of bass note reproduction would be more perceivable than the same small frequency error at HF?

In psychoacoustics, it is known that up to about 500Hz (AFAIR) sound location is perceived using ITD (InterAural Timing differences) & above, it changes to ILD (loudness differences) - does better close-in jitter explain this?